Linux operating systemsAIX operating systems

Manage GPFS file system data with storage pools

Edit online

With Global Parallel File Systems (GPFS) technology, you can manage your data using storage pools. A storage pool is a collection of disks or RAIDs with similar properties that are managed together as a group.

The group under which the storage pools are managed together is the file system. The automated placement and management of files on the storage pool level is done by policies. A policy is a set of rules that describes the life cycle of user data, based on the attributes of the file.

When a file is created, the placement policy determines the initial location of the data of the file and assigns the file to a storage pool. All data written to that file is placed in the assigned storage pool. The management policy determines file management operation, such as migration and deletion. The files within a GPFS file system are distributed over different storage pools, depending on the enabled placement and migration policies.

During restore, the files are placed on the correct storage pool. The IBM® Storage Protect server is not aware of pool-to-pool migrations, so the files are placed on the storage pool from where the backup has taken place. The policy engine replaces the files based on migration policies.

If a storage pool ID is stored in the extended attributes of the file, and that storage pool is available, the file is always placed in that storage pool. If the storage pool is not available, the file is placed according to the placement policy. If the placement policy does not match the file, the file is placed in the system pool.

GPFS handles the placement of files after a restore as follows:

  • The file is placed in the pool that can be selected by matching the saved file attributes to a RESTORE rule
  • The file is placed in the pool that it was in when it was backed up
  • The file is placed based on the current placement policy
  • The file is placed in the system storage pool

The GPFS RESTORE rule allows you to match files against their saved attributes rather than the current file attributes. If the file attributes do not match, GPFS tries to restore the file in the sequence described above.

For more information about the GPFS RESTORE rule, read the GPFS documentation about policies and rules.

The following restrictions apply:

  • The restore of stub files does not work with multiple storage pools, or with files that have ACLs
  • Unlink of filesets are not allowed
  • The ctime option of GPFS should be set to no (default), to prevent unwanted Backup-Archive backups of files after GPFS file migration from pool to pool

For information about using storage pools, see the IBM Storage Protect server documentation.